The building sector has become the largest consumer of end use energy in the world, exceeding both the industry and the transportation sectors. Extensive types of energy saving techniques have been developed in the past two decades to mitigate the impact of buildings on the environment. Instead of the conventional active building environmental control approaches that solely rely on the mechanical air conditioning systems, increasing attention is given to the passive and mixed-mode approaches in buildings. This thesis aims to explore the integration of passive cooling approaches and active air conditioning approaches with different dehumidification features, by making effective use of the information on: 1) various dynamic response properties of the building system and mechanical plants, 2) diverse variations of the building boundary conditions over the whole operation process, 3) coupling effect and synergistic influence of the key operational parameters, and 4) numerous parameter conflicts in the integrated active-passive operation. These issues make the proposed integration a complex multifaceted process operation problem. In order to deal with these challenges, a systematic approach is developed by integrating a number of advanced building/system physical models and implementing well established advanced dynamic optimization algorithms. Firstly, a reduced-order model development and calibration framework is presented to generate differential-algebraic equations (DAE) based physical building models, by coupling with the high-order building energy simulations (i.e., EnergyPlus) and implementing MLE+ co-simulation programs in the Matlab platform. The reduced-order building model can describe the dynamic building thermal behaviors and address substantial time delay effects intrinsic in the building heat transfer and moisture migration. A calibration procedure is developed to balance the modelling complexity and the simulation accuracy. By making use of the advanced modeling and simulation features of EnergyPlus, the developed computational platform is able to handle real buildings with various geometric configurations, and offers the potential to cooperate with the dominant commercial building modeling software existing in the current AEC industry. Secondly, the physical model for the active air conditioning systems is developed, which is the other critical part for the dynamic optimization. By introducing and integrating a number of sub-models developed for specific building components, the model is able to specify the dynamic hygrothermal behavior and energy performance of the system under various operating conditions. Two representative air conditioning systems are investigated as the study cases: variable air volume systems (VAV) with mechanical dehumidification, and the desiccant wheel system (DW) with chemical dehumidification. The control variables and constraints representing the system operational characteristics are specified for the dynamic optimization. Thirdly, the integrated active-passive operations are formulated as dynamic optimization problems based on the above building and system physical models. The simultaneous collocation method is used in the solution algorithm to discretize the state and control variables, translating the optimization formulation into a nonlinear program (NLP). After collocation, the translated NLP problems for the daily integrated VAV/DW operation for a case zone have 1605/2181 variables, 1485/2037 equality constraints and 280/248 inequality constraints, respectively. It is found that IPOPT is able to provide the optimal solution within minutes using an 8-core 64-bit desktop, which illustrates the efficiency of the problem formulation. The case study results indicate that the approach can effectively improve the energy performance of the integrated active-passive operations, while maintaining acceptable indoor thermal comfort. Compared to the conventional local control strategies, the optimized strategies lead to remarkable energy saving percentages in different climate conditions: 29.77~48.76% for VAV and 27.85~41.33% for DW. The energy saving is contributed by the improvement of both the passive strategies (around 33%) and active strategies (around 67%). It is found that the thermal comfort constraint defined in the optimization also affects the energy saving. The total optimal energy consumption drops by around 3% if the value of the predicted percentage dissatisfied (PPD) limit is increased by one unit between 5~15%. It is also found that the fitted periodic weather data can lead to similar operation strategies in the dynamic optimization as the realistic data, and therefore can be a reasonable alternative when the more detailed realistic weather data is not available. The method described in the thesis can be generalized to supervise the operation design of building systems with different configurations.